World Heritage Sites are often described as places of Outstanding Universal Value — but how many people actually visit them each year? World Heritage Explorer looks at what visitor numbers reveal, where the data comes from, and what the figures can and cannot tell us about the global reach of heritage tourism.
The Most-Visited World Heritage Sites
The table below ranks World Heritage Sites by annual visitor numbers, using the most recent available count for each site.
Heritage Tourism Is Enormously Concentrated
The numbers confirm what anyone who has queued at the Acropolis or tried to find space in front of the Taj Mahal already knows: heritage tourism is profoundly unequal. The Historic Centre of Rome leads the list with around 22 million visitors in 2024; the Sun Temple at Konârak recorded 3.6 million in the same year; the memorial sites of the Rwandan genocide received fewer than 100,000 in their most recent count. At the very bottom of the 93 sites with any data at all, the Rock Drawings in Valcamonica recorded just 3,900 visitors in 2022.
This concentration has measurable economic consequences. A 2024 study by Kutlu and colleagues tracked international tourist arrivals across all 27 EU member states over two decades and found that countries with more UNESCO World Heritage Sites consistently attract more international visitors — each additional site in a country’s portfolio is associated with a small but reliable uptick in arrivals. Among all the cultural factors they examined, UNESCO site count had the largest effect.
But the same relationship that makes inscription valuable also reinforces existing inequalities. Countries that already have many sites attract more visitors, which funds better infrastructure and promotion, which attracts more visitors still.
The study also found that hotel capacity — the sheer number of beds available near a destination — was an even stronger driver of tourist arrivals than UNESCO status. This is intuitive once stated: the Acropolis benefits from Athens, Versailles benefits from Paris. Sites embedded in major cities have an enormous structural advantage over isolated heritage sites, however outstanding their universal value. A remote temple or a rural landscape simply cannot offer the hotels, restaurants, and connections that turn a day trip into a multi-day stay.
Research using social media as a proxy for visitor flows confirms the concentration pattern in finer detail. Falk and Hagsten’s 2021 study analyzed approximately 680 million Instagram posts tagged at around 525 World Heritage Sites in Europe and North America. Historic city sites received roughly 40 times more posts than any other site type. At the very top, Paris, Banks of the Seine accumulated over 115 million posts; Istanbul’s historic areas nearly 90 million; the Historic Centre of Rome over 53 million. At the other extreme, prehistoric and industrial heritage sites received the fewest posts, with some rural mining landscapes recording fewer than 100.
How the Top Sites Break Down
Several patterns emerge from the visitor table.
European historic cities dominate the extreme top. Rome (22.2M), Paris (20.3M), and Vienna (8.2M) lead the rankings, reflecting a combination of global brand recognition, dense tourism infrastructure, and the relative ease of measuring city-wide arrivals from hotel statistics.
US national parks account for a substantial cluster in the middle tier. Great Smoky Mountains (12.2M), Grand Canyon (4.9M), Yellowstone (4.7M), Yosemite (4.1M), and Olympic (3.7M) all appear in the top 15. The National Park Service’s systematic visitor monitoring — counting every vehicle or pedestrian entry — makes these among the most reliably documented sites in the world.
Iconic cultural monuments attract visitors specifically for themselves. Pompeii (4.6M), the Acropolis (4.6M), the Taj Mahal (6.5M in 2018), and Angkor (955,000 in 2025 — still well below its pre-pandemic peak of 2.2 million) are global pilgrimage destinations in a secular sense: people travel to them, not past them.
Memorial sites punch above their weight. Auschwitz-Birkenau received 1.83 million visitors in 2024, ranking 23rd globally among sites with data — a remarkable figure for a site defined by its obligation to bear witness. The Hiroshima Peace Memorial reached a record 2.26 million visitors in fiscal year 2024, its first year ever above two million, boosted by the 2023 G7 summit in the city and the awarding of the 2024 Nobel Peace Prize to Nihon Hidankyō, the Japanese atomic bomb survivors’ organisation. Independence Hall drew just over 3 million visitors as part of Independence National Historical Park in 2023, with the site’s 250th anniversary in 2026 expected to push numbers higher still.
One finding from Falk and Hagsten deserves particular attention: sites on the UNESCO World Heritage in Danger list attracted approximately four times more Instagram posts than comparable unlisted sites. The danger designation, intended to flag sites at risk, appears to amplify their visibility and appeal instead. The label meant to protect a site also functions as a powerful marketing signal — one of the built-in contradictions of the World Heritage system.
These concentrations matter not just for statistics but for site management. Overcrowding and physical wear are recurring challenges at the most-visited sites. The Acropolis limits daily access to 20,000 people; Pompeii introduced entry restrictions in late 2024; Versailles has begun discussing visitor caps.
Where Visitor Data Exists — and Where It Doesn’t
Of the 1,248 World Heritage Sites in the dataset, only 93 — fewer than one in thirteen — have any visitor figure recorded at all. The chart below shows how that data gap is distributed across different kinds of heritage sites (see WHE typology).
The picture is striking in every direction.
Protected Areas & National Parks are the best-documented category relative to their size: 5 sites appear in the global top 20, a further 15 have visitor data, and 145 have none. Systematic monitoring programs run by bodies like the US National Park Service or Parks Canada account for much of this coverage — but even here, the majority of inscribed protected areas worldwide go uncounted.
Archaeological Sites have the largest absolute data gap: 236 sites have no visitor figure, against only 17 that do. The 2 in the global top 20 — led by the 4.6 million visitors to the Acropolis — are celebrated global icons; the remaining 236 have no published count.
Historic Cities & Urban Areas present a paradox: 3 of the world’s 20 most-visited sites are historic cities, yet 201 cities in this category have no recorded figures. The cities with data tend to be among the most famous in the world; hundreds of lesser-known historic urban centres go uncounted.
Buildings & Architectural Ensembles show the broadest absolute coverage — 23 sites have data and 5 are in the top 20 — but 175 still have none.
Cultural Landscapes stand out for a different reason: not one of the 59 inscribed cultural landscapes has any recorded visitor figure whatsoever. These living working landscapes — agrarian territories, designed parks, managed coastlines — are among the most difficult to count, lacking gates, fences, or a single point of entry. Their complete absence from the data is partly a measurement problem and partly a reflection of lower priority in heritage tourism promotion.
Memorial Sites are a small category with modest documentation: 2 sites are in the global top 20 and 2 more have data, while 9 have none. The category includes some of the world’s most emotionally significant destinations — Auschwitz, Hiroshima, Robben Island, the Rwandan genocide sites — whose visitor numbers carry moral as well as statistical weight. The 96,300 visitors recorded at the Rwanda genocide memorial sites in 2017 stands in especially sharp contrast to the millions drawn to European and East Asian sites.
The pattern reinforces what Falk and Hagsten found with Instagram data: early-inscribed, well-known, and physically bounded sites dominate both the visitor rankings and the data record. Sites inscribed more recently, or those in less-visited categories, are systematically under-represented — both in tourist footfall and in the statistics that would allow their management.
Do Visitors Actually Know They Are at a World Heritage Site?
Visitor numbers tell us how many people come. They say nothing about what the World Heritage designation means to those visitors once they arrive.
Research suggests the answer is often: very little. King and Halpenny’s 2014 study surveyed 1,827 visitors at five World Heritage Sites in Queensland, Australia, and 712 visitors at Hawaii Volcanoes National Park. Across both settings, more than 96% of visitors could not recall what the World Heritage symbol represented. Awareness that the site was listed at all was higher — 60% of Queensland visitors knew — but fell to just 19% among Hawaii visitors.
The implications are significant. If the primary function of the World Heritage brand is to prompt visitors to value and protect a site, that function is largely failing. King and Halpenny found that management agencies took a laissez-faire approach to branding: the UNESCO symbol appeared on entrance signs, but no systematic effort was made to explain what it meant.
Large visitor numbers do not indicate that a site is being experienced as World Heritage. As Falk and Hagsten show, the sites that dominate the visitor rankings were globally celebrated before the UNESCO program began — their numbers reflect pre-existing fame, not any tourism-enhancing effect of inscription. And as Kutlu et al. find at the country level, the economic benefit of World Heritage listing operates mainly through the cumulative prestige of a national portfolio, not through awareness at any individual site. Individual site branding, it turns out, is a different matter from aggregate tourism effects.
What the Numbers Measure — and What They Do Not
Visitor counts for World Heritage Sites are not standardized. Different sites use different counting methods, and the figures in any comparative table reflect that inconsistency.
Some World Heritage Sites are inscribed as a collection of separate components — Paris, Banks of the Seine covers four distinct areas; the Medici Villas and Gardens in Tuscany spans multiple separate properties. Where a parent site has no direct visitor figure but its individual components do, the analysis uses Wikidata’s “has part” property (P527) to identify all components, fetches the most recent figure for each, and sums them.
Three broad methodological categories
Even setting aside coverage gaps, the underlying figures are not all measuring the same thing:
- Ticket-based counts record the number of paid admissions. This is the most precise method, used at enclosed sites such as Versailles, the Acropolis, and Pompeii.
- City-wide or regional estimates use hotel arrivals or transport statistics as proxies. Rome’s 22 million figure reflects ISTAT municipal hotel arrival data — not a count of people entering a specific heritage zone. These figures are methodologically incomparable with ticket-based counts.
- Social media proxies — Instagram posts, Flickr photographs, Wikipedia page views — are increasingly used in academic research where official counts are absent. Falk and Hagsten show that Instagram post counts correlate reliably with official figures where both exist, though their data cover Europe and North America only, and Instagram users skew younger, more educated, and higher-income than the general visitor population.
A simple ranking of visitor numbers should therefore be read with caution. A site near the top of the table because it uses city-wide hotel data is not necessarily more visited than one near the bottom that counts only ticket-holders.
Anyone Can Contribute
Because Wikidata is community-maintained, anyone can add or update visitor figures. If a site you know has published more recent data — an annual report, a government statistics release, or a press statement — the figure can be added to Wikidata under property ‘visitors per year’ (P1174) with the reference year as a ‘point in time’ (P585) qualifier. It will be picked up automatically the next time this analysis runs.
Data quality and temporal gaps
Beyond methodology, the data face a further challenge: uneven coverage in time. The most recent figures in the table range from 2005 to 2025, reflecting the uneven availability of published statistics worldwide.
The COVID-19 pandemic created a sharp break in many series. Angkor received 2.2 million visitors in 2019 but only 955,000 in 2025, still well below its pre-pandemic peak. Some entries are simply very old: the Australian Convict Sites and Tasmanian Wilderness figures both date to 2005.
By contrast, well-resourced sites publish regular, detailed reports. The Palace of Versailles produces an annual activity report; English Heritage publishes Stonehenge’s figures annually; Italy’s Ministry of Culture publishes standardized figures for all state-managed sites, making Italian heritage among the best-documented in the world.
This asymmetry in reporting capacity mirrors broader inequalities in heritage management resources. And it echoes Kutlu et al.’s finding that government cultural expenditure is itself a positive driver of tourist arrivals: countries that invest more in their cultural sector tend also to invest in measuring it. The data gap is not random — it correlates with the same structural inequalities that shape the visitor numbers themselves.
What the Numbers Tell Us
Four main conclusions emerge from the data and the research.
Heritage tourism is heavily concentrated. A handful of sites — mostly in Europe and North America, mostly urban, mostly inscribed in the early years of the program — account for a disproportionate share of all visits. Falk and Hagsten put this precisely: historical city sites attract around 40 times more social media attention than other site types, and early-inscribed sites consistently outperform recently listed equivalents.
UNESCO inscription has a measurable tourism effect — mainly at the country level. Kutlu et al.’s analysis of EU countries finds a clear 0.22% increase in international arrivals for every 1% increase in a country’s World Heritage Site count. But this aggregate effect does not translate into awareness at the individual site level: the brand that works for France as a whole is largely invisible to visitors standing in front of a specific monument.
The figures are not directly comparable. Ticket-based counts, city-wide hotel estimates, park-entry counters, and social media proxies all measure different things. Any ranking of World Heritage Sites by visitor numbers is an approximate overview, not a precise league table.
Quantity does not equal understanding. King and Halpenny’s finding that over 96% of surveyed visitors at major World Heritage Sites could not identify the UNESCO symbol is a direct challenge to the assumption that visitor numbers proxy for heritage engagement. A site can be simultaneously over-visited and under-understood — and closing that gap is as much a management challenge as managing the crowds themselves.
The data coverage gap — visible clearly in the typology chart above — is itself a finding. The sites we know most about are, almost by definition, the sites that already have the resources, the infrastructure, and the global profile to attract attention. The 236 archaeological sites, 201 historic cities, and 59 cultural landscapes with no visitor figures at all represent the largely invisible majority of the World Heritage list: inscribed, protected in principle, but uncounted.
Sources: Falk, M.T. & Hagsten, E. (2021). Visitor flows to World Heritage Sites in the era of Instagram. Journal of Sustainable Tourism, 29(10), 1547–1564. King, L.M. & Halpenny, E.A. (2014). Communicating the World Heritage brand: visitor awareness of UNESCO’s World Heritage symbol and the implications for sites, stakeholders and sustainable management. Journal of Sustainable Tourism, 22(5), 768–786. Kutlu, D., Zanbak, M., Soycan, S., Kasalak, M.A. & Aktaş Çimen, Z. (2024). The Influence of World Heritage Sites on Tourism Dynamics in the EU 27 Nations. Sustainability, 16, 9090. Visitor figures sourced from Wikidata (property P1174), retrieved March 2026.